Looking for a job

Smart jack

@smartjack

AI evaluation engineer at Handshake AI authoring containerised agent benchmarks with deterministic verifiers and release gates.

India

Message

At Handshake AI, I design containerised agent-evaluation benchmarks for frontier-model training in security, cryptography, and ML inference. I wrote deterministic verifiers and multi-test suites for a hostile-server task that was merged after maintainer review and Pass@3 model trials.

At Meva Agro Products (OPC) Pvt. Ltd., I built a QR, OCR, and vision-LLM pipeline to check more than 39,000 field photos against a master list, with zero wrong reads found in audits. I also built the operations data layer for attendance, payroll, labour-efficiency KPIs, and reporting.

At Lumivon AI Studio, I run an applied-AI studio delivering LLM-powered assistants and websites, and built a real-time Kashmiri/Urdu voice agent. I bring earlier experience in KYC/AML analysis, insurance operations, and retail, alongside an MBA.

Experience

Work history, roles, and key accomplishments

Current

AI Evaluation Task Author

Current

Handshake AI

Sep 2026 - Present (2 months)

Design containerised agent-evaluation benchmarks for frontier-model training, with reference oracles and no-op baselines. Wrote deterministic verifiers and multi-test suites, with tasks merged after maintainer review and Pass@3 model trials.

Pytest Docker Agent Evaluation Benchmark Authoring Security Cryptography ML Inference

Current

Founder & Lead Engineer

Current

Lumivon AI Studio

Jan 2026 - Present (10 months)

Run an applied-AI studio delivering LLM-powered assistants and websites. Built a real-time Kashmiri/Urdu voice agent and evaluated STT models by word error rate, diagnosing orthography mismatch as the main error source.

LLM Speech Recognition TTS N8n Supabase PostgreSQL Docker Security Audits

Current

Software Developer

Current

Meva Agro Products (OPC) Pvt. Ltd.

Jan 2023 - Present (3 years 10 months)

Led an ML data programme reading 39,000+ field photos with a QR to OCR to vision-LLM cascade, lifting verified coverage to 78% with zero wrong reads. Architected multi-agent automation with n8n, Claude Code and Codex, cutting operational turnaround by 40%.

Python OCR Vision LLMs N8n Appsheets Google Apps Script Kotlin Docker SQL

Store Manager

Nutraj

Feb 2023 - Dec 2023 (10 months)

Ran store operations and a team of 4, using sales and stock data to grow revenue 15% and cut stockouts 25% through replenishment controls.

Inventory Control Sales Analysis Team Management

KYC Executive

Concentrix

Jan 2020 - Jan 2021 (1 year)

Performed AML/CDD analysis including ultimate beneficial owner discovery, proof-of-funds review, and PEP screening, with zero compliance slippage on high-risk escalations.

KYC AML CDD PEP Screening Compliance

Process Associate

Tata Consultancy Services

Jan 2014 - Jan 2017 (3 years)

Managed policy lifecycle and automated quotation dispatch for UK insurer Phoenix Life, maintaining over 96% SLA accuracy.

Policy Management SLA Process Optimization

Education

Degrees, certifications, and relevant coursework

Mercor

Introduction to Agents · AI Agent Evaluation, AI Agent Evaluation

Completed AI Agent Evaluation certification with a Certificate of Recognition for rigor, calibration, and clarity in AI agent evaluation.

Dr. APJ Abdul Kalam Technical University

Master of Business Administration (MBA), Business Administration

Grade: CGPA 7.26

Earned a Master of Business Administration with a CGPA of 7.26. The project focused on ICICI Bank ratio analysis with EPS 10.56, NPM 12.33%, and ROCE 2.3.